Salience Induction against Multi-Hop RAG Agents: Threat and Defense

Abstract

Agentic retrieval-augmented generation (RAG) systems increasingly retrieve external evidence and orchestrate tools for knowledge-intensive applications. In Multi-Hop question answering, agents chain facts across documents. Existing defenses focus on content poisoning, which injects false facts, and prompt injection, which embeds directives. We identify a third attack surface: the salience channel, through which fact position, emphasis, framing, and semantic proximity can redirect reasoning even

Categories

Framework mappings

Suggested from the entry's categories.

@misc{zhou2026salience,
  title = {{Salience Induction against Multi-Hop RAG Agents: Threat and Defense}},
  author = {Xingfu Zhou and Pengfei Wang and Yuan Zhou and Wei Xie and Xu Zhou},
  year = {2026},
  month = jul,
  eprint = {2607.17535},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2607.17535}
}